Effective Spectral and Excitation Modeling Techniques for LSTM-RNN-Based Speech Synthesis Systems

Computer Science

• Long Short-Term Memory Network 100%
• Recurrent Neural Network 60%
• Deep Neural Network 40%
• Text-to-Speech 40%
• Modeling 40%
• Experimental Result 20%
• Perceptual Quality 20%
• Research Result 20%
• Statistical Parameter 20%
• Spectral Envelope 20%
• Speech Synthesis System 20%
• Filter Parameter 20%
• Synthesis Filter 20%
• Varying Nature 20%
• Linguistics 20%